Adaptive control of robot manipulator with radial-basis-function neural network

Abstract

Based on the inertia-related adaptive control scheme for a robot manipulator, a radial-basis-function neural network is included to compensate for the highly nonlinear system uncertainties. The adjustable parameters of the radial-basis-function neural network are adapted on-line according to an analytically derived learning algorithm. It is demonstrated by simulation that very fast convergence of the trajectory errors can be achieved even in the presence of the parametric and/or structural uncertainties in the manipulator model.